the robots are taking over

My last role was head of marketing for an AI personalisation company (building machine learning models for gambling operators and game studios), so thinking about how AI will impact marketing is a topic that I’ve dedicated a lot of thought to.

Marketing has largely been identified as one of the business functions where AI will have the most impact (if not the most), and fortunately, it is also one where tooling and integration is typically lower friction, and the options more sophisticated. We’ve all heard the ‘adapt or die’ rhetoric when it comes to AI, which is a bit sensational, but I think it does have some salience – if only to identify which companies prioritise innovation and are open to change, and which are not. It’s also transformative for small teams with limited budgets and resources. If you can wield your AI stack effectively (and with the advent of AI agents), it can feel like having a small army at your disposal.

We’ll see marketing teams will gradually shift their time and energy away from volume-based tasks, putting the onus on more creative areas that require human intervention - and I believe particularly in creative production. This will become the place where meaningful differentiation can be achieved at scale.

(1) Preparing teams for a shift in outputs

Arguably the most visible change we are starting to see now is AI bridging the gap between ‘good’ and ‘average’ marketing – particularly when it comes to the domains of content creation and campaign execution. Where I think is the ripest area for differentiation is in two distinct stages of marketing planning: (1) the foundational big-picture strategy; and (2) the thorough scrutiny and quality assurance of the outputs. A lot of volume-based marketing work ‘in the middle’ will see more and more automation benefits (efficiency), particularly with operational work involving data and mechanical tasks.

I would focus on these two areas; firstly, really invest in the fundamentals – ensuring that we prioritize a thoughtfully crafted, coherent, and clear marketing strategy – ultimately, a vision that we all stand behind (a hopeful prerequisite to this is that you already have product-market fit). This also ensures that we have good intent-driven inputs that benefit AI and any other outcomes.

Likewise, in areas where marketing creativity and strategy is concerned, human judgement is particularly crucial, given ongoing concerns about the accuracy of outputs and questions of ethics in AI. We need to sweat the details and be obsessed with meticulous detail in everything we do; this is what will make good marketing ‘great’. While increased volume is helpful, it should never compromise our commitment to quality. Good judgement, as usual, is notoriously tricky to teach - but that is often a separate question.

(2) Thoughtfully evaluating AI’s usefulness and role in the business

I’m not an advocate of blindly embracing AI. I believe I have a more measured approach to its adoption or use than many. When we look at AI applications from a marketing capability perspective, it should mirror our broader approach and ethos to AI at an organisational level, and it’s important we establish those parameters at the outset – around authenticity, accuracy, and principles.

One product-centric use case sees AI successful where it helps to simplify the user experience and make usability more intuitive (i.e. I find integrating natural language processing is usually helpful - although it often comes with engineering tradeoffs and costs). A similar example in marketing can arise around authenticity. Even if an image generation tool can create perfect and indiscernibly real photos of our team members, it doesn’t pass the test for me as it compromises our authenticity and the values we stand for.

A more nuanced case I’ve thought about is around writing, which we do a lot of in marketing. I often spar with AI to help me get through creative roadblocks, and iterate through awkward phrasing, but in the best case, I like to go about and tackle it myself first. When you start outsourcing your thinking and your thought processes that have served you well for so long, you risk losing a part of your identity and individuality - a piece of what makes you authentically “you” (and it’s usually very clear when someone has relied on ChatGPT to write for them).

AI is an enabler of existing strategy, not a standalone strategy. The way I approach AI and any new tech is that they are tools that present us with more options to help us succeed with our existing objectives and strategies. The best ones will seamlessly fit into an existing workflow.

The hard part (and where it is sometimes more art than science) is discerning which use cases serve to benefit the most from AI. This is highly variable and common pitfalls range from lack of sophistication of the technology, poor integration into your tech stack and processes, and enduring ways of working that are difficult to change.

It’s important to find the balance between preserving the old, while knowing when to introduce the ‘new’. Choosing what products to invest in is an area of judgement and will always come with some element of risk, but I think that in most cases these risks can be mitigated with proper due diligence, research, and low-stakes trialing, where you can fail safely.

(3) Futureproofing the team and investing in AI skills training

Given this, a crucial emerging skill is the ability to sift through and identify which tools actually work, and how to get the best out of them. In my experience, a lot of AI products are more form than substance, which will lead to non-starters before you can even evaluate fit and appropriateness.

AI has a place in every marketer’s tech stack to help them to do their existing job more effectively and efficiently, but even at the top of its game, they still require human intervention.

I’ve tested several tools and in the realm of generative AI. For example, I find Ideogram and Mindjourney works well for images, Dall-E does not; Claude has now overtaken ChatGPT-4 for me as the gold standard for text generation (though not without flaws… and too many outages!). I love using AI to criticise my work, and give me feedback on tone and style. For SEO-optimised writing, I like Frase. I haven’t yet found a video generation product convincing enough for commercial use, but options like Pika Art and Runway ML help with simple concepts. AI is also a great copilot for iterating creativity, but often is not reliable in producing the final result.

Productivity utilities that I’ve championed, like Otter.ai captures and summarizes key takeaways for meeting minutes, and has been transformative for my previous team, saving us plenty of time and energy.

An important aspect is ensuring everyone on the team feels empowered and has license to openly explore and test these new tools; many of which aren’t difficult to get started with. Experimenting with beginner-friendly tools is a simple way to build confidence with AI. Related, we need to democratize tech skills so that everyone can feel empowered to navigate machine learning and AI tools (furthermore, adjacent disciplines like data science and analytics, are increasingly becoming essential knowledge at a basic level). I would put a base level of AI training and upskilling as a personal development item for everyone. It may not be the most urgent of priorities, but is a requisite for our roadmap to futureproof both the business and the team.

We may end up changing a few of our behaviours and find new helpers in the form of AI copilots and agents, but ultimately, the fundamentals and principles of marketing still stand. Having this knowledge on the team will give us the confidence to move forward decisively and be well-prepared, especially as the marketing ecosystem continues to evolve.

We need to still apply the same core values of genuine care and humanity in our decision-making, with or without AI.